Epicor Kinetic + job shops
AI for Epicor Kinetic Job Shops: Quote History, Job Variance and Schedule Risk
Short answer
AI for an Epicor Kinetic job shop connects a private model to Kinetic's REST API and Business Activity Queries so estimators, schedulers and managers can ask about quote history, job variance and late subcontract operations in plain English. It respects Kinetic security, is read-only by default, and shows the query used for each answer.
- ERP
- Epicor Kinetic
- Industries
- Job Shops, Make-to-Order, Contract Manufacturing
- Written for
- Operations Manager
A make-to-order job shop on Epicor Kinetic works a different problem from a repetitive plant. Each quote is its own estimate, each job its own routing, and the shop's margin depends on whether estimating, scheduling and purchasing decisions made on a Tuesday hold up by the following month. Kinetic captures all of it: quotes, jobs, operations, labor, material and subcontract purchases.
The difficulty is the distance between capture and insight. Kinetic has dashboards, Business Activity Queries and, for those who use it, Epicor Data Discovery. They work well for known questions. They are less friendly to a manager who wants to ask why the last forty jobs for one customer ran over, or which jobs will miss their dates because a subcontract operation is late.
Kinetic is a powerful, configurable system, and a mature installation carries user-defined fields, business process methods, custom BAQs and Epicor Functions that the original implementer understood and that the current team may not. Questions about the data model end up queued behind one analyst or a consultant.
This page covers private AI for Kinetic in a job shop: the tables and APIs it uses, the use cases that tend to pay back, where Kinetic's own tools and Epicor's AI features are enough, and what to ask any vendor before connecting a model to production.
What usually gets in the way
The problems we hear most from operations manager teams running Epicor Kinetic.
Quote history is not used when quoting
Kinetic holds past quotes and the jobs they became, but estimators rarely have time to compare a new RFQ against similar past work, so repeat under-quoting continues.
Job variance is explained slowly
Costs by operation and material are on the job, but understanding why a job ran over, whether scrap, rework, extra setups or subcontract cost, means reading across several screens and notes.
Subcontract operations drive due dates
Heat treat, plating, grinding and other outside operations are tracked on purchase orders and subcontract operations, and late returns are often found when the downstream operation cannot start.
Scheduling relies on a few people
Finite capacity scheduling in Kinetic reflects the data entered, but the judgment about what to expedite and what to move is held by an experienced scheduler, and exceptions are numerous.
Customization hides the data model
User-defined fields, custom BAQs and BPM logic make each installation different, and few staff know where a particular figure comes from.
Where AI earns its place in Epicor Kinetic
Each use case names the ERP objects it reads or writes, so your ERP team can judge the integration effort before anyone commits budget.
Similar-quote lookup
An estimator describes a new RFQ and the assistant finds comparable past quotes and the jobs they became, with quoted versus actual hours, material and subcontract cost.
Touches: QuoteHed, QuoteDtl, QuoteOpr, JobHead, JobOper, JobMtl
Outcome: Gives estimators evidence to adjust a quote before it is sent, reducing repeat under-quoting on recurring work.
Job variance explainer
For a job or a group of jobs, the assistant summarizes cost and hours versus estimate by operation and material, noting scrap, rework and subcontract differences.
Touches: JobHead, JobOper, JobMtl, LaborDtl, PartTran, POs for subcontract
Outcome: Moves post-job review from a manual hunt to a short summary the manager can act on.
Late subcontract risk list
The assistant lists open subcontract operations and their purchase orders past or near promise date, and the downstream operations and customer due dates affected.
Touches: JobOper subcontract flags, PODetail, PORel, JobHead due dates, Vendor
Outcome: Replaces discovery at the machine with an early exception list, with follow-up emails drafted for the buyer.
Schedule exception triage
The assistant points to jobs unlikely to meet their dates and gives the cause, such as a material shortage, a capacity overload on a resource group or a late outside operation.
Touches: JobHead, JobOper, ResourceGroup, Resource, JobMtl, PartWhse
Outcome: Lets the scheduler work from causes instead of scanning the entire schedule.
Customer profitability by part family
Managers ask which customers or part families earn the best margin after setup and scrap, and see the jobs behind the figures.
Touches: InvcHead, InvcDtl, JobHead, QuoteHed, Customer
Outcome: Supports pricing and capacity decisions with evidence rather than impressions.
Customization documentation
The assistant reads BAQs, UD fields and BPM directives to answer what a field means and where a figure is calculated, turning tribal knowledge into a searchable answer.
Touches: BAQs, UD tables and columns, BPM method directives, Epicor Functions
Outcome: Reduces dependence on one analyst and shortens onboarding for new administrators.
Order and delivery status answers
Customer service asks where an order stands and what remains to be done, and gets an answer from live job and shipment status.
Touches: OrderHed, OrderDtl, OrderRel, JobHead, ShipHead
Outcome: Fewer calls to the floor, and more consistent answers to customers.
Reference architecture
The AI layer calls Kinetic's REST API v2 and published BAQs with a dedicated read-only API user, so answers respect Kinetic security and company scoping. The model runs on infrastructure you control, and every answer shows the BAQ or query behind it.
- 1
Kinetic connector
Uses REST v2 service endpoints and BAQs through an API key and a dedicated integration user with the minimum security groups needed. BAQs are preferred over raw SQL so Kinetic's security applies.
- 2
Semantic layer
Maps JobHead, JobOper, JobMtl, QuoteHed, OrderHed, PODetail and resource data to job shop terms, including UD fields and your own BAQs.
- 3
Model serving
An open-weight model such as Llama, Qwen or Mistral class served with vLLM or Ollama on customer GPUs or in a private tenant.
- 4
Retrieval and agents
Structured BAQ calls for numbers, retrieval over job and quote notes and attachments, and agents that draft vendor follow-ups and variance summaries for approval.
- 5
Governance and audit
Per-user logging mapped to the Kinetic user, with company and plant scoping respected.
Integration notes for your ERP team
- Prefer BAQs exposed through REST v2 over direct SQL, so Kinetic security, company and site scoping all apply.
- Create a dedicated integration user with narrow security groups and an API key, and rotate the key on your normal schedule.
- Inventory UD columns and custom BAQs early, since job shops often store material specs, certs and customer codes in UD fields.
- Rate-limit and schedule heavy calls; for large history analysis use a reporting replica or a scheduled export instead of live calls.
- Epicor's own AI and Data Discovery may already cover some questions; the private layer is for cross-record questions, documents and agents.
- No direct database writes: any action, such as a PO change, is a draft approved by a person and entered through Kinetic.
Deployment options
On-premises Kinetic with local AI
Shops running Kinetic on their own servers with customer NDAs or export-control constraints.
Model, connector and index run on the same network with no outbound path for ERP data.
Private tenant beside Kinetic cloud
Shops on Epicor's cloud who want AI under their own control.
A dedicated tenant calls the Kinetic REST API over an authenticated channel; prompts and answers stay within your tenant.
Hybrid
Groups with some plants on-prem and others in the cloud.
Each connector uses that site's own API user, with a common governance layer and no central copy of the databases.
Compliance and data control
How the architecture supports your obligations. Certification and accountability stay with your organisation; the design keeps the evidence straightforward.
Customer NDAs and drawing controls
A model on infrastructure you control means job, quote and customer data are not sent to a public AI provider, which is what most NDAs require.
ITAR and export-controlled work
Controlled jobs and parts can be excluded from retrieval, or the full system can run in an environment limited to authorized persons.
ISO 9001 and AS9100 record control
Answers cite source records and any suggested change flows through existing change control rather than being applied automatically.
SOC 2 expectations from larger customers
Per-user logging, role mapping to Kinetic security groups and no external data egress give concrete answers for a customer's vendor security review. Netray does not itself claim your certification.
Where Netray fits
How an engagement runs
Phase 1 . 2-3 weeks
Discovery
- -Kinetic version, hosting and customization inventory
- -BAQ and UD field review
- -Ranked use cases with estimating, scheduling and purchasing
- -Deployment recommendation
Phase 2 . 6-8 weeks
Pilot
- -Read-only connector via REST v2 and BAQs
- -Similar-quote lookup and variance explainer live
- -Accuracy check against known jobs
- -Logging and security group review
Phase 3 . 4-6 weeks
Production
- -Late subcontract and schedule exception lists rolled out
- -Role-based access
- -Backup plan
- -Admin runbook
Phase 4 . Ongoing
Improve
- -Customization documentation assistant
- -Additional plants or companies
- -Model updates
- -Quarterly accuracy and usage review
Questions to ask any vendor, including us
A short list that separates real Epicor Kinetic AI work from a chatbot demo.
- Do you call Kinetic through REST and BAQs so our security groups apply, or do you read the database directly?
- Where does the model run, and can we verify no job or quote data leaves our environment?
- How are UD fields and our custom BAQs handled?
- What load will the connector place on the Kinetic server or tenant, and how is it rate-limited?
- Can estimators see the jobs behind a suggested hours figure?
- Which questions are already answered by Kinetic dashboards, Data Discovery or Epicor's own AI?
- How are Kinetic version upgrades handled for the connector?
- What do we keep if the engagement ends?
Frequently asked questions
Does Epicor Kinetic's own AI make this unnecessary?
Sometimes. If Kinetic dashboards, Data Discovery or Epicor's AI features answer your questions, use them. A private layer helps with cross-record questions such as quote-to-job comparison, with documents and notes, and where data must stay on infrastructure you control.
How does it connect to Kinetic?
Through the REST v2 API and published BAQs using a dedicated read-only integration user and API key. Using BAQs means Kinetic's own security, company and site scoping apply, which is safer than reading the database directly.
Can it help estimators price better?
It surfaces evidence: comparable past quotes and the jobs they became, with quoted versus actual cost. The estimator still decides. Accuracy depends on how consistently jobs were costed and closed in Kinetic.
Does it write to Kinetic?
Not by default. It is read-only. It drafts vendor follow-ups and variance summaries, and anything entered in Kinetic is done by a named person through the normal process, so Kinetic business logic and BPMs still run.
What about our customizations?
They are inventoried in discovery: UD fields, custom BAQs and BPM directives. The semantic layer is mapped to them so questions resolve to your real data, and the assistant can also help document them.
Is our data sent to a public AI?
No. The model is open-weight and runs on infrastructure you own or in a private tenant, with no route to public AI APIs. Controlled jobs or customers can be excluded from retrieval.
How long does a pilot take?
Roughly six to eight weeks for a read-only pilot after two to three weeks of discovery, depending on API access approvals, customization complexity and available time from estimating and scheduling staff.
Related guides
AI for Epicor Kinetic, Beyond What Prism Covers
Add AI to Epicor Kinetic beyond Prism: private LLM over BAQs, BPM data, and REST v2, on-prem or private cloud, with honest guidance on when Prism already covers you.
Kinetic Cloud + private AIPrivate AI for Epicor Kinetic Cloud, Outside the Multi-Tenant Boundary
Ground a private LLM on Epicor Kinetic Cloud via BAQs, REST v2, and Data Fabric so sensitive manufacturing data never enters a multi-tenant AI feature.
Job shop ERP + AIAI for ECI JobBOSS2 and E2 Shop System
Add AI to ECI JobBOSS2 or E2 Shop System for faster quoting, tighter scheduling, and shop-floor answers, without sending your job data to the cloud.
Ask your ERP anythingNatural Language Query for ERP Data: Ask SAP, Infor, or Oracle a Question in Plain English
See how natural language query over SAP, Infor, Oracle, and NetSuite data works: grounded text-to-SQL, role-based permissions, and a visible audit trail.
RAG + SQL + permissionsA Private LLM Grounded on Your ERP Data
How a private LLM answers questions on your ERP data: RAG plus text-to-SQL, role-based permissions inherited from the ERP, and where each fits.
Agents + approval gatesAI Agents for ERP, Running On-Prem
A practical guide to on-prem AI agents for ERP: what they can safely automate, where human approval belongs, and how to design the guardrails.
Plan it with numbers
On-Prem AI ROI Calculator
Turn hours saved per employee into annual net benefit, payback months, and 3-year ROI for an on-prem AI investment.
Free ToolOpen-Weight Model Selector
A 10-question assessment that matches your hardware budget, workload complexity, and operational maturity to the right open-weight model size class.
Free ToolAir-Gapped AI Readiness Assessment
A 10-question assessment that scores how prepared your organization is to deploy and operate LLMs inside an air-gapped or classified enclave.
GuideAI-Optimized Production Scheduling for Manufacturing
Optimize production scheduling with AI. Reduce changeover time 30%, improve on-time delivery 20%, and maximize throughput using constraint-based optimization algorithms.
GuideAI-Powered Supplier Risk Assessment Tool
Assess supplier risk with AI-powered scoring. Monitor financial health, delivery performance, quality metrics, and geopolitical risk from ERP and external data sources.
GuideAI Configuration Review Agents for ERP Systems
Deploy AI agents to audit ERP configurations detecting misconfigurations, security gaps, performance bottlenecks, and deviations from best practices automatically.
Talk it through with an engineer who knows Epicor Kinetic
Bring one real question your team cannot answer from the ERP today. We will map the data path, the model, and where it runs, and tell you honestly if AI is the wrong tool for it.